Design criteria for experiments to measure the breakdown voltage of insulating gases in uniform electric fields
Round robin breakdown experiments in (quasi)uniform electric fields have been performed in 7 different laboratories for air, HFO1234ze(E)/N2 (20 %/80 %), and SF6 as example gases. The experiments are supported and discussed with an extended literature review of gas-physical phenomena and constructional setup parameters influencing the results. The goals are to investigate a) the influence of certain experimental parameters on the outcome and scatter of breakdown measurements and b) to define suitable and practical setups for breakdown experiments and evaluation methods for the main experiments of Cigré working group D1.67 investigating new non-SF6 insulating gases and gas mixtures. The final recommendation is to use polished plane-plane Rogowski electrodes with a spacing of 5 mm in a self-supporting PEEK frame, and an AC voltage rate of rise around 0.1 kV/s. The value of this study goes beyond the work of the working group D1.67 and serves as basic recommendation for all groups performing breakdown experiments in gaseous insulation systems.
CIGRE Working Group D1.67 has the task to “…investigate and summarize the dielectric properties and the practical insulation performance of new non-SF6 insulating gases and gas mixtures”. In this context, an extended literature review has been performed and revealed that the present state of knowledge is incomplete and insufficient to fulfil this task. The most popular (but only seemingly a simple) approach to determine the electric withstand strength of (novel) insulation gas mixtures is to perform breakdown experiments. The variety of experimental setups and techniques used for performing and evaluating breakdown experiments is huge, which impedes a meaningful comparison of results. This is a general problem for the community, not only for this Working Group, as experimental results from different groups cannot be compared. In principle, even two measurement campaigns with seemingly contradicting results could both be correct if the different experimental conditions are known and would be properly taken into account in the analysis. Thus, the Working Group members decided to carry out round robin experiments with a selection of new gas mixtures and clearly specified experimental setups and conditions. It was decided to start with a series of pre-tests with the goals to a) investigate the influence of certain experimental parameters on the outcome and scatter of breakdown measurements and b) to define suitable and practical setups for breakdown experiments and evaluation methods, which will then be used during a main round robin test series.
To the authors’ knowledge, currently one available standard describes a breakdown test method for insulating gases: ASTM D2477-07using a sphere-plane electrode arrangement with a sphere radius of 9.5 mm and 2.54 mm electrode spacing. Another available standard is IEC-60052 regarding voltage measurements in standard air gaps with sphere-sphere electrode arrangements with large radii. This standard is not for testing of different gases, but gives some recommendations about experimental conditions for proper voltage measurements in atmospheric air and provides reference values.
For the purposes a) and b), first a recapitulation of the theory and physics of gas breakdown is given. Here, the streamer criterion is applied to predict the breakdown voltage when the electron swarm parameters and the electric field distribution are known. In a next step, a detailed summary of a literature review is given and possible effects of experimental conditions on breakdown in compressed gases in (quasi) uniform fields are given. Their influence on the breakdown voltage is discussed in detail, together with an assessment of advantages and disadvantages with respect to experimental efforts. The discussed parameters are in particular: voltage wave shape, voltage application method, breakdown number, limitation of energy input by the breakdown discharge, electrode -geometry, -finish, - material, and - conditioning phenomena, artificial irradiation, pressure vessel-design, -material, as well as electrode spacing and gas pressure range.
Two basic phenomes are identified to influence the measured breakdown voltage: the statistical time lag which leads to an overestimation of the measured breakdown voltage, and electrode surface imperfections, which lead to an underestimation of the measured breakdown voltage. The statistical time lag can be reduced by increasing the availability of seed electrons by means of a large critical volume which is enclosed by electrodes, artificial irradiation, facilitation of cold field emission by electrode finish roughening, gas pressure increase, and lowering the rate of voltage increase. The influence of electrode surface imperfections can be reduced by means of electrode -polishing, -material with high evaporation temperature, -preconditioning, small critical volume, fast turn off time after breakdown, and low gas pressure.
Taking the conclusions from literature review into account, practical considerations and the available infrastructure and test capabilities of the involved laboratories, pre-tests focussing on breakdown in (quasi) uniform electric fields were performed in seven different labs. This pre-study was performed with example gases of air, HFO1234ze(E)/N2 (20 %/80 %), and SF6. Particular attention was paid to the results analysis to ensure high reproducibility, comparability in between different labs and minimum possible uncertainty. For this reason the parameters of electrode geometry, electrode finish, UV irradiation, rate of rise of applied AC voltage, and gas pressure were systematically varied.
Round-robin AC breakdown experiments were preformed, for polished and sandblasted stainless steel electrodes, with shapes of sphere-sphere (radii 30 mm and 37.5 mm ), sphere-plane (sphere radius 9.55 mm according to ASTM D2477-07), and plane-plane with Rogowski shaped edge (radius 35 mm and 60 mm), for electrode spacings in the range of 2.54 mm to 15 mm. For air in the pressure range of 0.1 MPa to 0.4 MPa, the results show a good agreement between laboratories for same experimental conditions, but large deviations if different experimental parameters as electrode shape, electrode surface, and voltage increasing rate were used. For this reason, in the next step, the experimental parameters in air were systematically varied. It was shown that polished sphere-plane and sphere-sphere arrangements in combination with fast voltage increasing rate of 5 kV/s result in the widest breakdown distribution ranges within the measurement series, with the highest breakdown values being up to two times larger than the lowest. This can result in up to 50 % scatter around the median. The presence of UV, lower voltage increasing rates, sandblasting, or replacement of the electrodes by plane-plane geometries significantly reduces the scattering range, to deviations from median down to a few percent. It is observed that this decrease of the distribution range results from the decrease of the level of the highest breakdown values. The level of the lowest breakdown values remains unchanged and closely corresponds to the calculated streamer criterion and breakdown values in reference literature. For new polished and sandblasted electrodes a trend of slightly increasing consecutive breakdown values is observed due to electrode conditioning effects in the early phase of the test series. For polished electrodes this trend is significantly shorter than for sandblasted. The same phenomena as in air are observed in SF6, in the pressure range of 0.02 MPa to 0.1 MPa using plane-plane electrodes. Above 0.1 MPa the breakdown values are below the calculated streamer criterion but above reference measurements available in literature. Mixtures of HFO1234ze/N2 20 %/80 % show a significant formation of black dust during the breakdown measurement series, leading to decreasing trends in the distribution throughout the measurement series. Calculations of statistical time lags with Volume Time Criterion show a good agreement with the distribution of the measured breakdowns. These time lags can be up to several tens of seconds.
For the experimental setup for measurements planned in the Working Group D1.67, the final recommendation is to use polished plane-plane Rogowski electrodes with a spacing of 5 mm in a self-supporting PEEK frame, and around 0.1 kV/s AC voltage rate of rise. The results should be evaluated without assumptions of a specific distribution, but calculation of median and 84.13- and 15.87- percentiles with 75 % of confidence.
The value of the reported parameter study goes beyond the work of the present Working Group and serves as basic recommendation for all groups performing or planning to perform breakdown experiments in gaseous insulation systems. Depending on the desired experimental boundary conditions and focus, other experimental conditions may have to be used. For this purpose, the influence of experimental conditions discussed and assessed in this work, should give ideas what has to be considered.
Working Group D1.67 has also published a Technical Brochure, summary available on this page.
